A Novel AI Program to Detect COVID-19 Pneumonia and the Distribution of Affected Loci

BackgroundChest computed tomography findings are beneficial for diagnosing COVID-19 pneumonia. However, few respiratory physicians and diagnostic radiologists can evaluate these computed tomography images. We investigated whether the artificial intelligence program InferRead™ CT Pneumonia is useful for detecting COVID-19 pneumonia and whether its affected volume distribution can aid in detecting COVID-19.MethodsFrom July 18 to August 17, 2021, chest computed tomography scans of 79 of 217 SARS-CoV-2-positive and 90 of 1094 negative patients were analyzed by the AI program. We evaluated the performance of the program in diagnosing SARS-CoV-2. We also calculated the volume of interest in each lung lobe.ResultsThe sensitivity and specificity of InferRead™ CT Pneumonia for the detection of SARS-CoV-2-positive status were 0.797 (95% CI: 0.692–0.880) and 0.644 (95% CI: 0.537–0.743), respectively. The agreement between the diagnosis of COVID-19 pneumonia by the program and that of two respiratory specialists was satisfactory high (respiratory specialist A: κ = 0.817, respiratory specialist B: κ = 0.890). On the other hand, in SARS-CoV-2 negative cases, the results showed low concordance. Multiple logistic regression analysis revealed that the odds ratio of age for the inconsistency in SARS-CoV-2 negative cases with specialist A and B was 1.030 (95% CI, 0.997–1.070; P = 0.077) and 1.060 (95% CI, 1.020–1.110; P = 0.007), respectively. The median volume of interest for the lower lung lobes was particularly high: 11.52% in the right lower lobe (IQR: 4.08–26.99) and 7.16% in the left lower lobe (IQR: 2.27–20.04). In contrast, the right middle lobe showed relatively low (0.16%, IQR: 0.00–5.04), and a significant difference between affected lobes was detected (P < 0.001).ConclusionsInferRead™ CT Pneumonia is useful for detecting COVID-19 pneumonia especially in SARS-CoV-2-positive cases, and its affected volume distribution can aid in detecting COVID-19 pneumonia.